The purpose of the present study was to analyze the composition of marine bacterial communities around the world, and to investigate bacterial isolates regarding the production of antibiotics. This included molecular analyses of marine bacterioplankton, as well as culture-based studies of marine bacterial isolates with antagonistic activity. The work was based on samples collected during the Galathea 3 and LOMROG-II marine research expeditions that have explored many different oceanic regions worldwide. A molecular survey of marine bacterioplankton at 24 worldwide stations investigated the abundance of major bacterial groups, potential biogeographical patterns, and their relation to environmental parameters. The original aim was to determine whether the composition of the total microbiota correlates with the occurrence of culturable bioactive bacteria. No such correlation was found. Quantitative community analyses showed latitudinal patterns in bacterial distribution, revealing significantly different relative abundances of Bacteroidetes, unclassified Bacteria and Vibrio between warmer and colder oceans. Absolute cell numbers of most bacterial groups were positively correlated with nutrient concentrations in warmer oceans, and negatively with oxygen saturation in colder oceans. The finding of differing communities in warmer and colder oceans underlined the presence of biogeographical patterns among marine bacteria and the influence of environmental parameters on bacterial distribution. Studies of antagonistic isolates focused on six bioactive Vibrionaceae isolated during Galathea 3. The six strains were identified as Vibrio coralliilyticus (two strains), V. neptunius (two strains), V. nigripulchritudo (one strain), and Photobacterium halotolerans (one strain) by sequencing of housekeeping genes. Chemical metabolite profiling underlined genetic relationships by showing highly similar production of secondary metabolites for each species. Two known antibiotics were purified; andrimid from V. coralliilyticus and holomycin from P. halotolerans. In addition, two novel cyclic peptides from P. halotolerans and a novel siderophore-like compound from V. nigripulchritudo were isolated. All three compounds interfere with quorum sensing in S. aureus. During LOMROG-II further seventeen strains with antagonistic activity were isolated, affiliating with the Actinobacteria (8 strains), Pseudoalteromonas (4 strains), the Vibrionaceae (3 strains), and Psychrobacter (2 strains). Seven of the eight bioactive Actinobacteria, being isolated from different sources throughout the Arctic Ocean, were related to Arthrobacter davidanieli. Its broad antibiotic spectrum was likely based on production of the known arthrobacilin antibiotics. The eighth actinomycete, tentatively identified as Brevibacterium sp., produces a potentially novel antimicrobial compound. Most studies of antagonistic marine bacteria have been conducted with the aim of isolating novel antimicrobials with potential clinical applications. However, little is known about production and role of these compounds in the natural environment. This thesis took one step in this direction and demonstrated that V. coralliilyticus S2052 produced its antibiotic andrimid when grown with chitin as the sole carbon source. Whilst the strain produced an array of secondary metabolites in laboratory media, it focused on andrimid production with chitin. This indicates that the antibiotic is likely produced in the natural habitat and may serve an ecophysiological function. The finding that two related strains from public culture collections do not produce andrimid and have different biosynthetic temperature optima suggested that V. coralliilyticus may comprise different subspecies with different niches. In summary, the present study shows biogeographical patterns of marine bacterioplankton on a global scale. In addition, the thesis work has demonstrated that marine Vibrionaceae and polar Actinobacteria are a resource of antibacterial compounds and may have potential for future natural product discovery.